Warm water and big winds: The 2017 North Atlantic hurricane season was highly active, with six major storms—nearly two standard deviations above the normal number. Three of those storms made landfall over the Gulf Coast and the Caribbean, causing terrible damage and loss. Why was the season so fierce? Murakami et al. used a suite of high-resolution model experiments to show that the main cause was pronounced warm sea surface conditions in the tropical North Atlantic. This effect was distinct from La Niña conditions in the Pacific Ocean that were involved in other years. It remains unclear how important anthropogenic forcing may be in causing such increased hurricane activity. Science , this issue p. 794

Although anthropogenic climate change has contributed to warmer ocean temperatures that are seemingly more favorable for Atlantic hurricane development, no major hurricanes made landfall in the United States between 2006 and 2016. The U.S., therefore, experienced a major hurricane landfall drought during those years. Using the high-resolution Geophysical Fluid Dynamics Laboratory 25 km grid High-Resolution Forecast-Oriented Low Ocean Resolution (HiFLOR) global climate model, the present study shows that increases in anthropogenic forcing, due to increases in greenhouse gasses, are associated with fewer long-duration major hurricane landfall droughts in the U.S., which implies an increase in major hurricane landfall frequency. We create six different fixed-distance ‘buffers’ that artificially circle the United States coastline in 100 km radial increments and can compensate for the bias in hurricane landfall calculations with six-hourly datasets. Major hurricane landfall frequencies are computed by applying the buffer zones to the six-hourly observed and simulated storm track datasets, which are then compared with the observed recorded major hurricane frequencies. We found that the major hurricane landfall frequencies generated with the 200 km buffer using the six-hourly observed best-track dataset are most correlated with the observed recorded major hurricane landfall frequencies. Using HiFLOR with an implemented buffer system, we found less frequent projections of long-duration major hurricane landfall drought events in controlled scenarios with greater anthropogenic global warming, which is independent on the radius of the coastal buffer. These results indicate an increase in U.S. major hurricane landfall frequencies with an increase in anthropogenic warming, which could pose a substantial threat to coastal communities in the U.S.

The slowdown of the Atlantic Meridional Overturning Circulation (AMOC) is ubiquitous in climate projections, representing a major risk of global warming with far-reaching climatic impacts. Here, we investigate how the future AMOC slowdown can affect the activity tropical cyclones (TCs). To that end, we compare two sets of global warming simulations: one that exhibits AMOC weakening and another in which AMOC intensity is fixed. Using these experiments, we compute Genesis Potential Indices (GPI), to assess large-scale ocean-atmosphere conditions for TC formation, and conduct downscaling TC simulations for the two scenarios. Our analysis shows strongly enhanced tropical cyclogenesis in the North Atlantic, especially along the U.S. eastern seaboard and in the Gulf of Mexico, in a warm climate with a weakened AMOC. The AMOC slowdown causes roughly 60% of the estimated seasonal increase of 12 storms in the Atlantic with warming. Higher TC potential intensity (PI) in the North Atlantic due to greater air-sea thermodynamic disequilibrium and, to a lesser extent, reduced vertical windshear explain these findings, which highlight the important role of AMOC slowdown in 21st -century TCs.

This study investigates the relative contributions of large-scale thermodynamic and dynamic processes to decadal and multidecadal changes in Atlantic tropical cyclone (TC) activity, spanning the historical record since the late nineteenth century, and extending to 2100 projections. We employ a framework that decomposes TC counts into precursor disturbances that transition into fully developed storms, applied to multiensemble simulations of two TC-permitting atmospheric models. Using these models, we conduct controlled experiments with distinct SST forcings that isolate the influence of SST spatial patterns from that of global-mean warming on Atlantic TC activity. Our results show that decadal and multidecadal changes in TC frequency are primarily governed by two thermodynamic variables: potential intensity and moist entropy deficit. In the historical record, these variables reinforced one another, producing more robust trends in TC activity. In contrast, future projections suggest opposing influences, with one variable (potential intensity) becoming more favorable for TCs while the other (moist entropy deficit) becomes less favorable, leading to increased uncertainty in TC projections. We trace this shift to differences in relative warming between the tropical Atlantic and the broader tropics, underscoring that regional SST patterns, rather than the global-mean warming rate, control both past variability and projected future changes in TC activity. Constraining future projected patterns of warming is therefore essential for improving the reliability of TC projections.

The 2020 Atlantic tropical cyclone (TC) season was unprecedented, producing a record storm count that exceeded the seasonal forecasted ranges, despite their anticipation of high activity. Here, we assess the likelihood of the extreme 2020 season given the observed sea surface temperature (SST) forcing by examining a hierarchy of statistical, dynamical, and deep learning (DL) modeling frameworks, all constrained by observed SSTs. Although physics-based models anticipated a moderately active season, the observed outcome lay beyond their ensemble ranges. To investigate this result, we first test if the 2020 large-scale environment was conducive to a hyperactive season. Although the 2020 conditions were relatively favorable for TCs, they were not indicative of potentially record-breaking TC activity. We also generate 1000 samples of plausible 2020 Atlantic TC season outcomes given the observed SSTs by leveraging a DL model. From this large ensemble, the observed season emerges as a 0.5% event, which is highly unlikely, but not implausible. These results suggest that SST forcing provided a limited seasonal signal and that internal atmospheric variability could have driven the observed hyperactivity on top of the moderately favorable large-scale conditions. The inability of physics-based model ensembles to encompass the observed outcome does not indicate a model failure but reflects both limited predictability from SST forcing and the inability of relatively small ensembles [ N = O (10)] to sample and estimate extreme tail risk like the 2020 season, seasons which are extremely unlikely any given year but have a substantial probability of occurring some year across decades. Significance Statement: The 2020 Atlantic tropical cyclone (TC) season was the most active on record by number of named storms. Although forecasts anticipated above-normal activity, they did not encompass the observed hyperactivity. Even state-of-the-art physics-based, statistical, and deep learning (DL) models, when given “perfect” observed sea surface temperature (SST) inputs, did not capture the observed outcome within their simulated ranges. We assess how predictable this extreme season was from the observed SSTs. We show that while the large-scale conditions in 2020 were generally favorable for TCs, they were not exceptionally conducive to a record-breaking season. Using a 1000-member DL ensemble forced with observed SSTs, we find that the 2020 season was a rare outcome that a small ensemble could not have represented.

Seasonal Atlantic tropical cyclone (TC) counts are commonly modeled as a conditional Poisson process, implying that the distribution of possible seasonal outcomes—the range of TC counts that could plausibly occur in a given year—exhibits equidispersion for a given climate state, with its variance equal to its mean. This assumption underlies many statistical frameworks used for seasonal TC prediction and risk assessment, yet, to the best of our knowledge, has not been extensively tested directly. Using large ensembles from physics-based climate models (AM2.5-C360 and HiRAM) and a deep learning–based climate emulator [Ai2 Climate Emulator version 2 (ACE2)], we examine the distributional properties of seasonal TC counts in a controlled modeling framework. Across all models and years, we find that the ensemble distribution of within-season TC counts is systematically underdispersed relative to the Poisson distribution, with ensemble variances smaller than their corresponding means. This result provides evidence against modeling the SST-conditioned Atlantic TC count distribution with a Poisson distribution and instead supports the use of the binomial distribution. This representation of the Atlantic TC counts implies the following physical interpretation of TC genesis: TCs develop in two stages: 1) first from a finite number of seed disturbances and 2) followed by a seed–TC transition that is governed by large-scale environmental conditions. We fit seasonal TC count distributions from a 1000-member ACE2 ensemble to the Poisson and binomial distributions and find statistical evidence that the binomial formulation provides a better fit to TC count distributions. The Poisson assumption is not supported by current climate model simulations, given that simulated seasonal TC counts are more constrained than implied by traditional Poisson-based frameworks. Significance Statement: Seasonal Atlantic tropical cyclone (TC) forecasts are critical for risk preparedness, energy planning, and insurance decision-making. Many statistical models used for these forecasts assume that TCs occur randomly, following a Poisson process. Leveraging unprecedentedly large ensembles from state-of-the-art physics-based and deep learning climate models, we show that simulated seasonal TC counts systematically violate this assumption. Instead, TC counts are more constrained. This behavior is consistent with a physical picture in which TCs arise from a finite number of precursor disturbances (“seeds”) whose likelihood of development depends on the large-scale environment. These results motivate more physically grounded statistical models for seasonal TC prediction.